Program Report
Recent and planned improvements to Consumer Price Indexes
Excerpt from Program Report:
Recent and planned improvements to Consumer Price Indexes
The Bureau of Labor Statistics is engaged in a continuing effort to improve the Consumer Price Index program. The focus of these efforts in recent years has centered on three areas: (1) employing more current spending patterns and samples of priced items; (2) reflecting the effects of consumer substitution in response to relative price change; and (3) accounting more completely both for changes in the quality of existing consumer goods and services and for new product introductions.
Updating expenditure patterns
The CPI has a two-tiered weighting structure; at the upper level the weights are based upon data from the Consumer Expenditure Survey. Historically, these expenditure weights were updated about every 10 years. The consumer expenditure weights in the CPI now are being updated at 2-year intervals; expenditure weights for the 1999�00 period were introduced with the release of data for January 2002. The 1999�00 expenditure weights replaced the weights for the 1993� period that had been introduced in January 1998. These new weights were 2 years old at the time of their introduction in January 2002. In contrast, the previous, 1993�, expenditure weights were 3� years old when first used. Expenditure weights will next be updated to the 2001� period effective with release of data for January 2004. Although the expenditure weights are now being updated at 2-year intervals, the CPI抯 geographic area and housing unit samples are not being updated on a 2-year basis. However, the feasibility of a program for continuous rotation of these samples is being studied.
Updating items and outlets
The samples of items priced for the CPI and the retail stores and service establishments in which they are collected now are being updated on a 4-year cycle. Previously, these samples were updated every 5 years. In addition, because of improvements to the survey from which these samples are obtained, it now is possible to focus updating efforts on those areas of consumer spending in which goods and services tend to change rapidly.
Another major enhancement will occur in 2003, when BLS will begin reselecting many CPI item samples within the existing outlet samples, midway between the scheduled 4-year outlet sample rotations. That is, these item samples will be rotated every 2 years. This is a further step in making the CPI more representative of current consumer spending patterns as well as more reflective of new goods and services in the marketplace.
Accounting for consumer substitution
As noted, the CPI has a two-tiered weighting structure. Data from the survey of consumer expenditures do not provide enough detail to weight and combine prices at the individual item level. The weights for these items are presently derived from a survey of where consumers shop梒alled the Telephone Point-of-Purchase Survey. Within the two-tiered structure of calculating the CPI, consumer substitution can and does occur at both levels梩hat is, within and across item categories. To better approximate the effect of consumers� responses to changes in relative prices at the lower level of aggregation, BLS introduced a geometric mean estimator for averaging prices within most of the 211 index item categories.
The geometric mean is consistent with the assumption that consumers maintain constant expenditure shares at the elementary aggregate level. That is, it assumes that consumers substitute within item-area categories such that any increase in price of an item is offset by a corresponding decline in quantity purchased, and thus the item抯 expenditure share remains constant. The arithmetic mean, by contrast, is consistent with the assumption that consumers purchase fixed quantities of items within an item category and do not substitute in response to changes in relative prices.
The new formula now is being used in categories that, at the time of its introduction in February 1999, comprised approximately 61 percent of total consumer spending represented by the CPI-U. The remaining 15 index categories continue to be calculated as they previously had been, using an arithmetic mean. The geometric mean formula was chosen for those categories within which consumers are likely to change their spending in response to changes in relative prices. The arithmetic mean formula was chosen for those categories within which consumers are unlikely to respond to changes in relative prices.
Communications regarding the Monthly Labor Review may be sent to the Editor-in-Chief at 2 Massachusetts Avenue NE, Room 2850, Washington, DC, 20212, or faxed to (202) 691–7890.
Within Monthly Labor Review Online:
Welcome | Current
Issue | Index | Subscribe
| Archives
Exit Monthly Labor Review Online:
BLS Home | Publications & Research
Papers
